CN106662534A - Method for creating grade discrimination standard in granular object appearance grade discrimination device - Google Patents
Method for creating grade discrimination standard in granular object appearance grade discrimination device Download PDFInfo
- Publication number
- CN106662534A CN106662534A CN201580029634.2A CN201580029634A CN106662534A CN 106662534 A CN106662534 A CN 106662534A CN 201580029634 A CN201580029634 A CN 201580029634A CN 106662534 A CN106662534 A CN 106662534A
- Authority
- CN
- China
- Prior art keywords
- grade
- shot
- mentioned
- particle
- grade discrimination
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/28—Measuring arrangements characterised by the use of optical techniques for measuring areas
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/85—Investigating moving fluids or granular solids
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G06V10/7747—Organisation of the process, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
- G06V10/993—Evaluation of the quality of the acquired pattern
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
- G06V20/698—Matching; Classification
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/85—Investigating moving fluids or granular solids
- G01N2021/8592—Grain or other flowing solid samples
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30128—Food products
Abstract
This method for creating a grade discrimination standard in a granular object appearance grade discrimination device is provided with: a granular object mounting step for classifying a plurality of granular objects according to grades and mounting the granular objects on an image capture surface of an image capturing means; a captured image data acquisition step for capturing images of the granular objects mounted on the image capture surface by the image capturing means and acquiring captured image data; a grade information acquisition step for acquiring grade information of the granular objects according to the grades on the basis of the captured image data; and a grade discrimination standard creation step for creating a grade discrimination standard using the grade information acquired according to the grades.
Description
Technical field
The present invention relates to the grade discrimination base of a kind of grain of shot-like particle appearance ratings discriminating gear or particle equigranular thing
Accurate generation method.
Background technology
A kind of known grain appearance ratings discriminating gear (referring for example to patent document 1), it passes through scanner etc. and shoots single
Unit shoots multiple grain and obtains photographed data, and according to the photographed data grade of above-mentioned multiple grain is differentiated.
Above-mentioned grain appearance ratings discriminating gear shoots the multiple grain for becoming grade discrimination object simultaneously by shooting unit
Photographed data is obtained, class information (outer shape, area, length, width, the color of grain are obtained according to above-mentioned photographed data
(RGB information), crackle etc.), and the class information of above-mentioned grain is compared to sentence with the grade discrimination benchmark being previously set
The grade of not above-mentioned grain.
Also, the grade of multiple grain can easily and rapidly be differentiated according to above-mentioned grain appearance ratings discriminating gear.
But, in existing grain appearance ratings discriminating gear, the photographed data of multiple sample grains, operation are obtained in advance
Person passes through range estimation and confirms above-mentioned photographed data over the display while specifying the grade of each sample grain successively, thus using root
The class information of the above-mentioned each sample grain obtained according to above-mentioned photographed data is generating grade discrimination benchmark.
But, operator as described above needs expertly to confirm above-mentioned photographed data over the display on one side while specified etc.
Level, therefore it is very difficult that the grade discrimination benchmark of shot-like particle is generated in grain appearance ratings discriminating gear.
Prior art literature
Patent document
Patent document 1：Japanese Unexamined Patent Publication 2011-242284 publications
The content of the invention
Problems to be solved by the invention
Therefore, even if it is an object of the invention to provide a kind of not to be that skilled operator also can simply generate granular
The generation method of the grade discrimination benchmark in the shot-like particle appearance ratings discriminating gear of the grade discrimination benchmark of thing.
For the method for solve problem
In order to achieve the above object, the generation side of the grade discrimination benchmark of shot-like particle appearance ratings discriminating gear of the invention
Method, shoots shot-like particle and obtains photographed data by shooting unit, and according to above-mentioned photographed data the grade of shot-like particle, the party are differentiated
Method possesses：By multiple shot-like particles according to the shot-like particle mounting step being positioned in after grade separation on the shooting face of shooting unit；It is logical
Cross above-mentioned shooting unit and shoot the above-mentioned shot-like particle being positioned on above-mentioned shooting face and the photographed data acquirement for obtaining photographed data
Step；The class information for obtaining the class information of above-mentioned shot-like particle according to above-mentioned grade according to above-mentioned photographed data obtains step
Suddenly；And generated using the above-mentioned class information obtained according to above-mentioned grade grade discrimination benchmark grade discrimination benchmark generate
Step.
It is above-mentioned that the generation method of the grade discrimination benchmark of above-mentioned shot-like particle appearance ratings discriminating gear can also possess registration
The class name register step of the class name of shot-like particle.
Pass through above-mentioned class information in the generation method of the grade discrimination benchmark of above-mentioned shot-like particle appearance ratings discriminating gear
The class information that acquisition step is obtained can be included in outer shape, area, length, width, color and the crackle of shot-like particle
At least one.
In above-mentioned shot-like particle mounting step, the multiple shot-like particles being positioned on the shooting face of above-mentioned shooting unit are housed
It is being spaced to be classified in the pallet in multiple regions and according to grade.
In above-mentioned grade discrimination benchmark generation step, above-mentioned grade discrimination base can be generated using Machine Learning algorithms
It is accurate.
The effect of invention
In the generation method of the grade discrimination benchmark of the shot-like particle appearance ratings discriminating gear of the present invention, possessing will be multiple
Shot-like particle according to the shot-like particle mounting step being placed on after grade separation on the shooting face of shooting unit, can process in advance according to
The shot-like particle of grade separation, even if therefore not being grade discrimination base that skilled operator can also simply generate shot-like particle
It is accurate.
In addition, by the way that operator is placed on the shooting face of shooting unit in advance by the shot-like particle of level sorting, calculating
Machine can automatically generate grade discrimination benchmark.
Further, can reuse in advance by the shot-like particle of grade separation, therefore, it is possible to repeatedly generate with objectivity
Grade discrimination benchmark.
If above-mentioned grade discrimination benchmark generation step generates above-mentioned grade discrimination benchmark using Machine Learning algorithms, can
Enough simply generate high-precision grade discrimination benchmark.
Description of the drawings
Fig. 1 is the explanation of that represents the shot-like particle appearance ratings discriminating gear used by the method for the present invention
Figure.
Fig. 2 is the flow chart of the generation step of grade discrimination benchmark.
Fig. 3 is the explanatory diagram of situation about being placed on shot-like particle on the shooting face of shooting unit.
Fig. 4 is the flow chart of the generation example of grade discrimination benchmark.
Fig. 5 is the flow chart of grade discrimination step.
Fig. 6 is the flow chart of grade discrimination example.
Specific embodiment
Embodiments of the present invention are illustrated with reference to the accompanying drawings.
[structure of shot-like particle appearance ratings discriminating gear]
Fig. 1 represents of the shot-like particle appearance ratings discriminating gear used by an embodiment of the invention.
Shot-like particle appearance ratings discriminating gear 1 possesses and shoots rice, wheat, soybean, the grain such as corn or particle equigranular thing
Shooting unit 2, the computer 4 being connected with the shooting unit 2 by cable 3.City can be for example used in above-mentioned shooting unit 2
The scanner or compounding machine of field sale.
The cover that above-mentioned shooting unit 2 possesses body 2a, the shooting face 2b being located above body 2a, switchs shooting face 2b
2c.In addition, the light accepting part that above-mentioned body 2a possesses light source and is made up of colored CCD line sensor etc., above-mentioned light source is by by illumination
It is mapped to the composition such as white fluorescent lamp, White LED of shot-like particle of mounting on above-mentioned shooting face 2b, above-mentioned colored CCD line sensing
Device receives the reflected light from the shot-like particle.Also, above-mentioned shot-like particle G is housed in pallet 5, and it is positioned in above-mentioned shooting
On the 2b of face.
Above computer 4 possesses：Image processing part, it is by the shooting of the shot-like particle photographed by above-mentioned shooting unit 2
Data carry out image procossing, and extract the various grade letters such as the shape informations such as the optical information such as the color of the shot-like particle and profile
Breath；Grade discrimination benchmark generating unit, it generates grade and sentences using the above-mentioned class information extracted in the image processing part
Other benchmark；Grade discrimination benchmark storage part, it stores the grade discrimination benchmark of the generation；Grade discrimination portion, it is using being stored in
Above-mentioned grade discrimination benchmark in the grade discrimination benchmark storage part is differentiating the grade of above-mentioned shot-like particle；And display, its
The result that display is obtained by the grade discrimination portion.
Above-mentioned shot-like particle appearance ratings discriminating gear 1 is by the shooting of the multiple shot-like particles obtained by above-mentioned shooting unit 2
Signal is sent to above computer 4, and the grade of each above-mentioned shot-like particle is differentiated in the computer 4.
Hereinafter, the generation method of the grade discrimination benchmark in shot-like particle appearance ratings discriminating gear is illustrated.
[generation step of grade discrimination benchmark]
Fig. 2 represents the flow process of the generation step of the grade discrimination benchmark in shot-like particle appearance ratings discriminating gear.
(1) step S1：
The shooting unit 2 shown in Fig. 1 can be used in shooting.If the sample grain of each grade to be classified of operator is for example
The grain of rice, then in advance prepare whole grain, immature kernel, particle, cracked kernel, the sample grain for color tablets, foreign matter etc. and be grouped (A, B,
C ...), according to being placed on the shooting face 2b of shooting unit 2 after each component class these sample grains.Here, each above-mentioned group
The quantity of sample grain can be one, but be preferably many to obtain class information described later in a large number.In addition, it is not necessary that
Prepare the sample grain of equal number between group.
As shown in figure 3, when above-mentioned sample grain is placed on the shooting face 2b of above-mentioned shooting unit 2, by being received into
Be divided on a pallet 5 in multiple regions and can according to above-mentioned group (A, B, C ...) classified.Alternatively, it is also possible to
One pallet is used to a group, pallet for loading each group on the 2b of face respectively is being shot, it is also possible to can not readiness tray
In the case of etc., the region according to each group mounting is set on shooting face 2, in order to the sample grain do not organized with other mixes, and
Directly sample grain is placed on the 2b of shooting face.
In addition, above-mentioned sample grain needs to be classified according to each group, but need not arrange and be loaded.
(2) step S2：
Shoot in step S1 according to each said components class and the sample that is placed on the shooting face 2b of shooting unit 2
Grain, obtains the photographed data of each sample grain.
(3) step S3:
According to the photographed data obtained in step S2, computer 4 obtains the class information of each sample grain according to each group
(characteristic information) (class information acquirement).
Here, the outer shape of above-mentioned class information including sample grain, area, length, width, color (RGB information), split
Line etc..
(4) step S4：
The photographed data for confirming to be obtained in step S2 by display, it is excellent product/bad that operator specifies each group
Which (excellent finished product/bad products information is specified) of product.
Here, in the present embodiment, above-mentioned excellent finished product/bad products information is specified and is not required.If carry out
Above-mentioned excellent finished product/bad products specify, then in the grade discrimination of shot-like particle described later, except according to step S5 registration
The shot-like particle of grade discrimination object is differentiated to the class name in computer 4, it is also possible to be categorized as excellent finished product/bad products.
(5) step S5：
The title (class name) of above-mentioned each group is registered (input) in computer 4 (class name setting) by operator.So,
Class name and class information, excellent finished product/bad products information association are got up (grade separation information setting).Here, incite somebody to action etc.
Level differentiates that the sample grain of object is set to the grain of rice, is set as whole grain, immature kernel, particle, cracked kernel, 6 kinds of color tablets and foreign matter etc.
Level.
(6) step S6：
In the grade discrimination of shot-like particle described later, in the case where the grade equivalent to more than 2 is determined as, in order to sentence
A grade, setting priority Wei be only equivalent to (preferential differentiation order sets).On the other hand, do not set above-mentioned preferential suitable
Sequence can also be determined as corresponding all of grade.If entering row major differentiation order to set, a sample grain is counted as one
Individual grade, if do not enter row major differentiation order set, a sample is counted as corresponding all of grade.
(7) step S7：
Grade discrimination benchmark is generated using the above-mentioned class information according to grade separation.In the present embodiment, calculate
Machine 4 is automatically generated grade and is sentenced using cluster analysis and the analysis of self-adaptive enhancement algorithm (AdaBoost (Machine Learning algorithms))
Other benchmark.With regard to cluster analysis, referring for example to Japanese Unexamined Patent Publication 2010-60389 etc..With regard to self-adaptive enhancement algorithm, referring for example to
Japanese Unexamined Patent Publication 2013-33331 etc..
<Grade discrimination benchmark generates example>
Illustrate that grade discrimination benchmark generates example.
Fig. 4 is that the grade discrimination benchmark of step S7 is generated, and expression has used cluster analysis and self adaptation to strengthen
The product process of the grade discrimination benchmark of the analysis of algorithm.
(a) cluster analysis
(1) step S10：
When cluster analysis is carried out, the threshold value and goal systems quantity used by the analysis is obtained.Above-mentioned threshold value with
And goal systems quantity is stored in advance in the memory of computer 4.Each sample grain classification is being drawn by cluster analysis
Above-mentioned threshold value is used when being divided into system, it is whether correct in order to judge the distance between the system calculated according to the class information of input
Also above-mentioned threshold value is used.Above-mentioned goal systems quantity is the desired value of the system quantity that classifying and dividing is carried out by cluster analysis.
(2) step S11：
From in the class name that step S5 is registered select a grade, using step S3 obtain class information to this with
Outer grade is that the sample grain of non-selection grade carries out cluster analysis.The analysis is the analysis for carrying out self-adaptive enhancement algorithm
Pretreatment, it is different from the class name setting that the operator of step S5 is carried out, and with the sample of the above-mentioned non-selection grade of classifying and dividing
For the purpose of the situation of product grain.By using the classifying and dividing that the cluster analysis is carried out, operator in step sl can be excluded and existed
Prepare the impact of division mistake produced during the sample grain of each grade etc. in advance.
(3) step S12：
By the classifying and dividing of the cluster analysis of step S11, it is confirmed whether to be classified and is divided into above-mentioned goal systems quantity.
In the present embodiment, goal systems number is set into 6.This is experimental field to obtain the system quantity for differentiating that result is outstanding.
When result, system quantity when the classifying and dividing of the cluster analysis of step S11 are different from goal systems quantity, change
Above-mentioned threshold value simultaneously repeats cluster analysis till consistent with above-mentioned goal systems quantity.System quantity is according to above-mentioned threshold value
Size increase and decrease, so automatically change threshold value and till being repeatedly analyzed until being divided into goal systems quantity.
(4) step S13：
Preserve by the sample grain classifying and dividing of non-selection grade for goal systems quantity result.For example, if in step
The grade that S11 is selected is whole grain, then the grade that the sample grain beyond whole grain and operator are registered in step S5 is independently
It is classified and is divided into 6 systems.In step S13, these systems are preserved with each grade.
(5) step S14：
The cluster analysis of step S11 is carried out in all of grade.In the present embodiment, 6 are categorized as by step S5
Grade, so carrying out cluster analysis according to these all of each grades, according to each grade system is obtained.
The analysis of (b) self-adaptive enhancement algorithm
(6) step S15：
In step S15, a grade in the registration of step S5 is selected, by the sample grain of the grade selected and divided
Class is divided into the sample grain of the system of in the system under the grade (in present embodiment 6), using taking in step S3
Class information carrying out the analysis by self-adaptive enhancement algorithm, calculate the threshold value/positive and negative/weights system of each class information
Number.In the present embodiment, the threshold value of above-mentioned each class information/positive and negative/weight coefficient becomes grade discrimination benchmark.
(7) step S16：
In step s 16, the threshold value/positive and negative/weights of each class information calculated by the analysis meter of step S15 are preserved
The analysis results such as coefficient.
(8) step S17：
Being confirmed whether the sample grain of the grade by all of system to selecting in step S15 has been carried out by self adaptation
Strengthen the analysis of algorithm.The analysis by self-adaptive enhancement algorithm is carried out according to each system.
(9) step S18：
In step S18, it is confirmed whether to have carried out the analysis by self-adaptive enhancement algorithm with all of grade.With all
The grade analysis that carries out by self-adaptive enhancement algorithm according to each grade, to all of grade by all of system-computed
The threshold value of each class information/positive and negative/weight coefficient, saves as grade discrimination benchmark, thus terminates the life of grade discrimination benchmark
Terminate into i.e. study.
[grade discrimination step]
Fig. 5 represents the flow process of the grade discrimination step of the shot-like particle of shot-like particle appearance ratings discriminating gear.
(1) step S21：
Multiple shot-like particles of grade discrimination object are placed on the shooting face 2b of shooting unit 2.
(2) step S22：
Shoot and be placed into the shot-like particle on the shooting face 2b of shooting unit 2 in step S21, obtain the shooting of each shot-like particle
Data.
(3) step S23：
The class information of each shot-like particle is obtained by computer 4 according to the photographed data obtained in step S22.The class information
It is identical with step S3.
(4) step S24：
Using the class information obtained in step S23 and the grade discrimination benchmark being generated in advance, strengthened by self adaptation and calculated
What method was carried out analyzes to differentiate the grade of each shot-like particle.
(5) step S25：
According to the differentiation result and the excellent product/bad products information of step S4 of step S24, computer 4 will be each granular
Which of excellent product or bad products thing be categorized as and simultaneously preserve data.
(6) step S26：
Preserve the grade discrimination result of step S24.
(7) step S27：
According to the classification results of step S25, shown by appropriate display and be classified as excellent product or bad products
Each shot-like particle.
(8) step S28：
According to the grade discrimination result of step S24, shown according to each grade by appropriate display after being differentiated
Each shot-like particle.
<Grade discrimination example>
Fig. 6 is of the grade discrimination based on the grade discrimination benchmark of step 24, represents grade discrimination flow process.
(1) step S31：
First, a grade is selected in step S31.
(2) step S32：
In the grade selected, the system that desired value (in present embodiment 6) are obtained by cluster analysis, so choosing
The system for selecting one of them.
(3) step S33：
The grade for alternatively going out and the threshold value of each class information of the self-adaptive enhancement algorithm of system/positive and negative/power
The analysis results such as value coefficient, from the result that memory reading is preserved in step s 16.
(4) step S34：
A shot-like particle is selected from the photographed data obtained by shooting.
(5) step S35：
Select the class information of a shot-like particle with regard to selecting.
The following is the analysis carried out by self-adaptive enhancement algorithm.
(6) step S36：
With regard to the shot-like particle selected in step S34, judge the class information selected in step S35 whether than in step
The threshold value that S33 reads in is big.
(7) step S37 and step S38：
When above-mentioned class information is bigger than above-mentioned threshold value, confirmed by step S37 positive and negative.On the contrary when above-mentioned class information ratio
Above-mentioned threshold value hour, is confirmed positive and negative by step S38.Aggregate value is obtained by above-mentioned positive and negative confirmation result.
Here, the positive and negative positive side and the feelings positioned at minus side for being to be located in above-mentioned class information above-mentioned threshold value of present embodiment
Under condition, the benchmark (mark) equivalent to the grade selected in the case of which side is judged.
(8) step S39：
By step S39 by the aggregate value obtained in step S37 or step S38 and the degree of belief obtained according to each system
(or similar degree) is added.
(9) step S40：
It is confirmed whether the operation that step S36～step S39 has been carried out for all of class information.It is all to what is used
Class information carries out the judgement of step S36.
(10) step S41：
It is confirmed whether the operation that step S36～step S39 has been carried out for all of shot-like particle.It is all of to what is photographed
Shot-like particle carries out the judgement of step S36.
(11) step S42：
It is confirmed whether the operation that step S36～step S39 has been carried out for all of system.The institute of the grade to selecting
There is system to carry out the judgement of step S36.
(12) step S43：
It is confirmed whether the operation that step S36～step S39 whether has been carried out for all of grade.To setting in step 5
All grades carry out the judgement of step S36.
(13) step S44：
Each shot-like particle according to photographing obtains multiple degree of beliefs, each granular to differentiate according to the size of the degree of belief etc.
The grade of thing.Thus the differentiation of grade is terminated.
According to the embodiment of the present invention, it is placed on after multiple sample grains are classified according to each group (each grade)
On the shooting face 2b of shooting unit 2, the sample grain being grouped in advance can be processed, even if therefore not being that skilled operator also can
Enough simply generate the grade discrimination benchmark of shot-like particle.
In addition, according to the embodiment of the present invention, sample grain that operator is grouped in advance according to each component class is simultaneously carried
Put on the shooting face 2b of shooting unit 2, thus computer 4 can automatically generate grade discrimination benchmark.
Further, according to the generation method of the grade discrimination benchmark of embodiment of the present invention, can reuse and press in advance
According to the shot-like particle that each group is divided, therefore, it is possible to repeatedly generate the grade discrimination benchmark of objectivity.
Embodiments of the present invention are the analyses that carry out using cluster analysis and by self-adaptive enhancement algorithm generating
Above-mentioned grade discrimination benchmark, therefore, it is possible to generate high-precision grade discrimination benchmark.
In addition, in the embodiment of the invention described above, the cluster analysis used in the generation of grade discrimination benchmark with
And the analysis carried out by self-adaptive enhancement algorithm, but can use the analysis that carried out by other Machine Learning algorithms with
Other known methods are generating above-mentioned grade discrimination benchmark.
The invention is not restricted to above-mentioned embodiment and its structure can be suitably changed in the range of without departing from invention.
Industrial workability
Even if it is not that skilled operator can also simply generate shot-like particle grade discrimination benchmark with regard to the present invention, therefore
With very high use value.
The explanation of reference
1：Shot-like particle appearance ratings discriminating gear, 2：Shooting unit (scanner), 2a：Body, 2b：Shooting face, 2c：Cover,
3：Cable, 4：Computer, 5：Pallet.
Claims (5)
1. a kind of generation method of the grade discrimination benchmark of shot-like particle appearance ratings discriminating gear, shoots granular by shooting unit
Thing simultaneously obtains photographed data, and according to above-mentioned photographed data the grade of shot-like particle is differentiated, it is characterised in that
The method possesses：
By multiple shot-like particles according to the shot-like particle mounting step being positioned in after grade separation on the shooting face of shooting unit；
The above-mentioned shot-like particle being positioned on above-mentioned shooting face and the shooting number for obtaining photographed data are shot by above-mentioned shooting unit
According to acquisition step；
The class information acquisition step of the class information of above-mentioned shot-like particle is obtained according to above-mentioned grade according to above-mentioned photographed data；
And
The grade discrimination benchmark that grade discrimination benchmark is generated using the above-mentioned class information obtained according to above-mentioned grade generates step
Suddenly.
2. the generation method of the grade discrimination benchmark of shot-like particle appearance ratings discriminating gear according to claim 1, it is special
Levy and be,
The method is also equipped with the class name register step of the class name for registering above-mentioned shot-like particle.
3. the generation method of the grade discrimination benchmark of shot-like particle appearance ratings discriminating gear according to claim 1 and 2, its
It is characterised by,
The class information obtained by above-mentioned class information acquisition step includes outer shape, area, length, the width of shot-like particle
At least one of degree, color and crackle.
4. the grade discrimination benchmark of the shot-like particle appearance ratings discriminating gear described in any one in claims 1 to 3
Generation method, it is characterised in that
Above-mentioned shot-like particle mounting step in, by the multiple shot-like particles being positioned on the shooting face of above-mentioned shooting unit be housed in by
Separate and classified in the pallet for multiple regions and according to grade.
5. the grade discrimination benchmark of the shot-like particle appearance ratings discriminating gear described in any one in Claims 1 to 4
Generation method, it is characterised in that
In above-mentioned grade discrimination benchmark generation step, above-mentioned grade discrimination benchmark is generated using Machine Learning algorithms.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2014117106A JP6481263B2 (en) | 2014-06-05 | 2014-06-05 | Preparation method of quality discrimination standard in granular material appearance quality discrimination device |
| JP2014-117106 | 2014-06-05 | ||
| PCT/JP2015/065922 WO2015186708A1 (en) | 2014-06-05 | 2015-06-02 | Method for creating grade discrimination standard in granular object appearance grade discrimination device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106662534A true CN106662534A (en) | 2017-05-10 |
Family
ID=54766777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201580029634.2A Pending CN106662534A (en) | 2014-06-05 | 2015-06-02 | Method for creating grade discrimination standard in granular object appearance grade discrimination device |
Country Status (7)
| Country | Link |
|---|---|
| US (1) | US10386307B2 (en) |
| JP (1) | JP6481263B2 (en) |
| KR (1) | KR102382399B1 (en) |
| CN (1) | CN106662534A (en) |
| BR (1) | BR112016028473B1 (en) |
| TW (1) | TWI681182B (en) |
| WO (1) | WO2015186708A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107862230A (en) * | 2017-11-14 | 2018-03-30 | 安徽虚境科技有限公司 | A kind of grade determines and input method |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6560707B2 (en) * | 2017-04-20 | 2019-08-14 | ファナック株式会社 | Machined surface quality evaluation device |
| JP7023180B2 (en) * | 2018-05-10 | 2022-02-21 | 大阪瓦斯株式会社 | Sake rice analyzer |
| CN109539996B (en) * | 2018-12-12 | 2021-06-04 | 安徽省农业科学院水稻研究所 | Method for measuring length of rice stigma |
| CN110044269A (en) * | 2019-04-08 | 2019-07-23 | 安徽省农业科学院水稻研究所 | A kind of method of the best Proper Sampling Period of rice stigma linear measure longimetry |
| JP7345776B2 (en) * | 2019-06-11 | 2023-09-19 | 近江度量衡株式会社 | Agricultural product sorting equipment |
| WO2022080504A1 (en) * | 2020-10-16 | 2022-04-21 | 国立研究開発法人農業・食品産業技術総合研究機構 | Determination device, learning device, determination method, learning method, and control program |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102159938A (en) * | 2008-10-09 | 2011-08-17 | 株式会社佐竹 | Tray for image reading device |
| CN102782481A (en) * | 2010-05-19 | 2012-11-14 | 株式会社佐竹 | Method for calculating weight ratio by quality in grain appearance quality discrimination device |
| WO2013145873A1 (en) * | 2012-03-27 | 2013-10-03 | 株式会社サタケ | Optical type granule sorting machine |
| WO2014002636A1 (en) * | 2012-06-27 | 2014-01-03 | 株式会社サタケ | Device for measuring appearance of grains |
| US20140147015A1 (en) * | 2012-11-26 | 2014-05-29 | Frito-Lay North America, Inc. | Method for Scoring and Controlling Quality of Food Products in a Dynamic Production Line |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IL82037A0 (en) * | 1987-03-29 | 1987-10-20 | Kalman Peleg | Method and apparatus for automatically inspecting and classifying different objects |
| JP2000146847A (en) * | 1998-11-16 | 2000-05-26 | Kett Electric Laboratory | Grain judgment device using neural network using clustering and voronoi division |
| JP2002312762A (en) * | 2001-04-12 | 2002-10-25 | Seirei Ind Co Ltd | Grain sorting apparatus utilizing neural network |
| TWI220011B (en) * | 2003-02-10 | 2004-08-01 | Sheng Shiuan Corp | Mold for producing upper stop of nylon safety zipper |
| JP2010060389A (en) | 2008-09-02 | 2010-03-18 | Horiba Ltd | Particle analyzer, data analyzer, x-ray analyzer, particle analysis method and computer program |
| JP5565869B2 (en) | 2011-08-01 | 2014-08-06 | 株式会社ユニバーサルエンターテインメント | Identification device and identification program |
-
2014
- 2014-06-05 JP JP2014117106A patent/JP6481263B2/en active Active
-
2015
- 2015-06-02 WO PCT/JP2015/065922 patent/WO2015186708A1/en active Application Filing
- 2015-06-02 KR KR1020167035637A patent/KR102382399B1/en active IP Right Grant
- 2015-06-02 BR BR112016028473-9A patent/BR112016028473B1/en active IP Right Grant
- 2015-06-02 US US15/315,427 patent/US10386307B2/en active Active
- 2015-06-02 CN CN201580029634.2A patent/CN106662534A/en active Pending
- 2015-06-04 TW TW104118053A patent/TWI681182B/en active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102159938A (en) * | 2008-10-09 | 2011-08-17 | 株式会社佐竹 | Tray for image reading device |
| CN102782481A (en) * | 2010-05-19 | 2012-11-14 | 株式会社佐竹 | Method for calculating weight ratio by quality in grain appearance quality discrimination device |
| WO2013145873A1 (en) * | 2012-03-27 | 2013-10-03 | 株式会社サタケ | Optical type granule sorting machine |
| WO2014002636A1 (en) * | 2012-06-27 | 2014-01-03 | 株式会社サタケ | Device for measuring appearance of grains |
| US20140147015A1 (en) * | 2012-11-26 | 2014-05-29 | Frito-Lay North America, Inc. | Method for Scoring and Controlling Quality of Food Products in a Dynamic Production Line |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107862230A (en) * | 2017-11-14 | 2018-03-30 | 安徽虚境科技有限公司 | A kind of grade determines and input method |
Also Published As
| Publication number | Publication date |
|---|---|
| JP6481263B2 (en) | 2019-03-13 |
| WO2015186708A1 (en) | 2015-12-10 |
| BR112016028473B1 (en) | 2021-08-17 |
| US20170199132A1 (en) | 2017-07-13 |
| JP2015230265A (en) | 2015-12-21 |
| US10386307B2 (en) | 2019-08-20 |
| TW201546440A (en) | 2015-12-16 |
| TWI681182B (en) | 2020-01-01 |
| KR20170015338A (en) | 2017-02-08 |
| BR112016028473A8 (en) | 2020-10-20 |
| KR102382399B1 (en) | 2022-04-01 |
| BR112016028473A2 (en) | 2017-08-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106662534A (en) | Method for creating grade discrimination standard in granular object appearance grade discrimination device | |
| JP5533245B2 (en) | Method for calculating weight ratio by quality in grain appearance quality discrimination device | |
| KR102103048B1 (en) | Optical granular substance sorter | |
| JP5163985B2 (en) | Granular product type inspection device | |
| EP2586873A1 (en) | Method for detecting microorganisms, device for detecting microorganisms and program | |
| WO2014087689A1 (en) | Image processing device, image processing system, and program | |
| JP6490365B2 (en) | MICROBIOLOGICAL TEST DEVICE VERIFICATION METHOD, MICROBIOLOGICAL TEST DEVICE VERIFICATION DEVICE AND PROGRAM | |
| US20160321825A1 (en) | Measuring apparatus, system, and program | |
| US20120067952A1 (en) | Code Quality Evaluating Apparatus | |
| CN105391931A (en) | Image processing apparatus, and method for processing image | |
| Çınar et al. | Determination of effective and specific physical features of rice varieties by computer vision in exterior quality inspection | |
| KR102037560B1 (en) | Device for measuring appearance of grains | |
| EP1324283A1 (en) | Document authenticity discriminating apparatus and method therefor | |
| JP2007080087A (en) | Facial parts extraction method and face authentication device | |
| Aznan et al. | Rice seed varieties identification based on extracted colour features using image processing and artificial neural network (ANN) | |
| US9449255B2 (en) | Method and apparatus for the determination of classification parameters for the classification of bank notes | |
| CN107389511B (en) | Crop seed phenotype measuring method and portable device | |
| JP6884584B2 (en) | Image processing device, imaging device, image processing method, image processing program, and storage medium | |
| CN112236668A (en) | Grain gloss measuring device | |
| JP6122655B2 (en) | Agricultural crop selection apparatus and crop selection method | |
| CN109543696A (en) | A kind of image-recognizing method neural network based and its application | |
| JPH0755688A (en) | Device for classifying and displaying particle | |
| CN112581539B (en) | Gray scale adjusting method, device, color selector and computer readable storage medium | |
| Malyadri et al. | Quality Testing of Rice Grains using SVM | |
| Jaisharma | RICE QUALITY ANALYSIS USING MACHINE LEARNING |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170510 |